Using pilot-symbol assisted modulation, we develop a method to acquire channel information from signals received in low-SNR environments. First, we estimate the power-delay profile (PDP) of the channel. Then, we apply information from that estimate to obtain the channel coefficients only at points where the PDP is nonzero. Because we estimate a reduced number of channel coefficients, the complexity and estimation error are simultaneously reduced. We present a case study of the proposed estimation method when applied to a wideband skywave high frequency (HF) channel. Through this study, we find that the proposed approach has significantly enhanced performance when compared to an existing method that does incorporate sparsity information into its estimate. The improved low-SNR performance of the proposed estimator makes it particularly well suited for spread-spectrum and underlay waveforms.